Nitrogen Catabolite Repression of DAL80 Expression Depends on the Relative Levels of Gat1p and Ure2p Production in Saccharomyces cerevisiae

GATA family activators (Gln3p and Gat1p) and repressors (Dal80p and Deh1p) regulate nitrogen catabolite repression (NCR)-sensitive transcription in Saccharomyces cerevisiae presumably via their competitive binding to the GATA sequences upstream of NCR-sensitive genes. Ure2p, which is not a GATA family member, inhibits Gln3p/Gat1p from functioning in the presence of good nitrogen sources. We show that NCR-sensitive DAL80 transcription can be influenced by the relative levels of GAT1 and URE2 expression. NCR, normally observed with ammonia or glutamine, is severely dimin-ished when Gat1p is overproduced, and this inhibition is overcome by simultaneously increasing URE2 expression. Further, overproduction of Ure2p nearly elimi-nates NCR-sensitive transcription under derepressive growth conditions, i.e. with proline as the sole nitrogen source. Enhanced green fluorescent protein-Gat1p is nuclear when Gat1p-dependent transcription is high and cytoplasmic when it is inhibited by overproduction of Ure2p. are

Success of the regulatory mechanism hypothesized above requires the relative amounts of Gln3p, Gat1p, Dal80-p, and Deh1p to be rigorously and finely controlled. Such fine control has been proposed to be achieved through autogenous and cross-regulation of GATA factor production (6,11), a view supported by (i) the presence of multiple UAS NTR elements situated upstream of the DAL80, DEH1, and GAT1 genes and (ii) effects of gln3⌬, gat1⌬, dal80::hisG, and deh1⌬ mutations on expression of the three genes (3,11,12). Although the model fits existing data, its central tenant has not been tested experimentally, i.e. that expression of one GATA factor, GAT1, influences expression of another, DAL80.
The purpose of this work was to test the above proposal by determining the effects of varying GAT1 expression on that of DAL80. The data obtained demonstrate that DAL80 expression is tightly linked to that of GAT1 and furthermore that NCR can be tightly linked to the relative production of Gat1p and Ure2p, i.e. that changing the relative levels of GAT1 and URE2 expression concomitantly changes the sensitivity of DAL80 expression to NCR. We also find that the intracellular localization of EGFP-Gat1p is influenced by the levels of GAT1 and URE2 expression. When Gat1p-dependent transcription is high, EGFP-Gat1p is predominantly nuclear, whereas when it is inhibited by overproduction of Ure2p, Gat1p is cytoplasmic.
Fluorescence Microscopy-Cytological experiments were performed as described elsewhere (25). The distribution of fluorescent material was the same whether cells were grown in Wickerham's or YNB minimal medium.

GAT1 Expression under Control of the GAL1,10 Promoter-
Dissecting the regulatory relationships of the GATA transcription factors is hampered by their extensive autogenous-and cross-regulation (11,12). To overcome this, we replaced the genomic GAT1 promoter with a GAL1,10 fragment (UAS GAL ), thereby removing GAT1 expression from nitrogen and GATA factor control and placing it under galactose regulation. Because our experiments depended upon the transcription profile of the GAL1,10-GAT1 construct, we measured steady-state levels of GAT1-specific mRNA following three different glucose additions to the medium. GAT1 mRNA increased as glucose decreased ( Fig. 2A). Next, we substituted the lacZ open reading frame for the GAT1 open reading frame in CEN-based GAL1,10-GAT1 pTSC666, measured ␤-galactosidase production under conditions similar to those in Fig. 2A, and found ␤-galactosidase activity decreased as the amount of glucose added to the medium increased (Fig. 2B). The inverse of the glucose addition was plotted on the abscissa throughout this work to facilitate visualization of the effects of GAL1,10 driven GAT1 or URE2 expression. As expected, GAL1,10-GAT1 expression is NCR-insensitive (Fig. 2B). In fact, GAL1,10-GAT1-lacZ expression is 2-and 4-fold higher, respectively, with glutamine and ammonia compared with proline. This effect probably derives from ammonia and glutamine being significantly better nitrogen sources than proline and hence supporting higher synthetic capacities (26).
DAL80 Expression Is Strongly Influenced by the Level of GAT1 Expression-A predicted consequence of the cross-regulation model of GATA factor control (11) is that the level of GAT1 expression influences DAL80 expression. We tested this prediction using a DAL80-lacZ reporter. We recognized from the outset that interpretation of our results could be compromised by: (i) down-regulation of DAL80-lacZ expression by Dal80p derived from the genomic DAL80 gene responding to increased Gat1p (6) and (ii) secondary effects of growing cells with galactose as carbon source. We eliminated the first complication by performing our experiments in dal80::hisG disruption strain TCY48. To evaluate the second potential complication, we compared ␤-galactosidase production in wild type (TCY5) and dal80::hisG disruption (TCY29) strains transformed using DAL80-lacZ pTSC572. The only difference in the DAL80-lacZ expression profile was a 20 -30% decrease with galactose as carbon source (Fig. 2C), a relatively modest effect relative to that of NCR.
With the experimental system established, we determined the effect of varying GAT1 expression on DAL80-lacZ expression in TCY48. As the glucose concentration decreased (GAT1 expression increased), DAL80-lacZ expression increased (Fig.  3). In addition to the DAL80 expression expected when proline was provided as nitrogen source, increasing GAT1 expression unexpectedly increased DAL80-lacZ expression with ammonia and glutamine as nitrogen sources as well (Fig. 3). To place this result in perspective, DAL80-lacZ expression with ammonia as nitrogen source in Fig. 3 (3,969 units) is about 300-fold greater than normal (glutamine, 12 units) (11). In short, increasing GAT1 expression overcomes the repressive effects of preferred nitrogen sources on DAL80-lacZ expression.
Increased URE2 Expression Suppresses the Effects of High Level GAT1 Expression-The current model of NCR proposes that the binding of Ure2p to Gln3p inhibits its ability to function as a transcriptional activator (27). Because Ure2p also regulates Gat1p-dependent NCR-sensitive transcription (9), we determined whether increasing URE2 and GAT1 expression simultaneously would inhibit the ability of GAT1 to suppress the NCR sensitivity of DAL80-lacZ expression. Because the GAL1,10-URE2 construct could not be made in exactly the same way as GAL1,10-GAT1, we first determined whether the overall GAL1,10-URE2 transcription profile was similar to that of GAL1,10-GAT1 (Fig. 2). Fig. 4 (A and B) shows that it was, except for a modest decrease in ␤-galactosidase production at the highest levels of URE2 expression; the reason for this decrease is unknown. We found that increasing URE2 and GAT1 expression together (Fig. 5A, solid lines) dramatically suppressed the ability of GAT1 overexpression (Fig. 5A, dotted lines) to support high level DAL80-lacZ expression with proline or ammonia as nitrogen source. Moreover, DAL80-lacZ expression with GAT1 and URE2 both overexpressed was fully NCRsensitive. These data show that high level URE2 expression neutralizes the effects of overexpressing GAT1. Furthermore, increasing GAL1,10-URE2 expression (in TCY57) abrogated the need of an excess nitrogen "signal," i.e. as URE2 expression increased in a culture provided with a nonrepressive nitrogen
Gat1p Appears in the Nucleus Only When There Is Gat1p-dependent Transcription-We previously showed a correlation between intracellular distribution of a GATA factor, Gln3p, and NCR-sensitive gene expression. 2 Therefore, we wanted to determine whether DAL80 expression correlated with the localization of Gat1p. We constructed ADH1-EGFP-GAT1 pRA27 and used it to transform gat1⌬ (RRJ715) cells. In this strain Ure2p is produced at wild type levels, and GAT1 is constitutively overexpressed. DAL80 was highly expressed in the transformants (Fig. 4C, lane B). When the experiment was repeated in strain RTCY57, which overproduces Ure2p, no DAL80 expression was detected (Fig. 4C), confirming the observations in Fig. 5A. When similarly prepared cultures were viewed micro-scopically, EGFP-Gat1p fluorescence was predominantly nuclear when URE2 was expressed at wild type levels (Fig. 6, K-Y) and co-localized with DAPI-stained material (panels a-i). On the other hand, no nuclear localization was observed when URE2 was overexpressed and DAL80 expression was inhibited (Fig. 4C). Fluorescence rather appeared in the cytoplasm as punctate spots (Fig. 6, A-J) similar to those observed when cultures were transformed with GFP-URE2 pNVS22 (data not shown, but it appears as in Fig. 9 of Cox et al. 2 and as reported earlier by Wickner's laboratory (28)).

DISCUSSION
The cross-regulation model (11) of GATA factor regulation in S. cerevisiae posits that GAT1 expression directly regulates the level of DAL80 expression and hence Dal80p production. To test this prediction we circumvented the complex normal crossregulation observed among GATA factor genes and their products by substituting GAL1,10 for the GAT1 and/or URE2 pro- FIG. 5. A, expression of the DAL80-lacZ gene in strains expressing GAT1 or GAT1 ϩ URE2 at various levels. Strain TCY48 was grown with either proline (E) or ammonium sulfate (ϫ) as sole nitrogen source in galactose minimal medium, to which was added the indicated amounts of glucose. Strain TCY60, containing plasmid pTSC572, was grown with either proline (Ⅺ) or ammonium sulfate (‚) as sole nitrogen source. B, DAL80-lacZ expression (from pTSC572) in strain TCY57 expressing URE2 at various levels. FIG. 4. GAL1,10-URE2 expression profile measured by Northern blot analysis (A) and by ␤-galactosidase production supported by GAL1,10-URE2-lacZ fusion (B). Details of the Northern blot analysis were as in Fig. 2 using strain TCY57 was grown in galactose-proline minimal medium with the indicated amount of glucose added. 32 P-Labeled probes were synthesized from the 0.45-kb BsrGI-ApaI fragment from URE2 and yeast pC4. ␤-Galactosidase production from pTSC668 (in strain TCY1) was measured as in Fig. 2. C, Northern blot analysis of total RNA (9 g/lane) from strain RRJ715, untransformed (lane A) or transformed pRA27 (lane B), and strain RTCY57 transformed with pRA27 (lane C). Strains were grown in minimal glucose ammonia medium, and RTCY57 was additionally transferred to galactose ammonia medium and induced for 3 h prior to assay. Full-length DAL80 and ACT1 probes were used. moters and demonstrated that we could obtain cultures containing differing amounts of GAT1 and URE2 expression and mRNA, which implies they also produced differing amounts of Gat1p and Ure2p. Using this experimental system, we demonstrated that DAL80-lacZ expression (i) depended upon GAT1 expression, (ii) became significantly insensitive to NCR when GAT1 was overexpressed, and (iii) regained NCR sensitivity when URE2 was simultaneously overexpressed with GAT1.
To further evaluate the relationship between GAT1 and DAL80 expression, we plotted GAL1,10-GAT1-lacZ expression data from Fig. 2B as a function of DAL80-lacZ expression (Fig.  3). A linear relationship exists at all but the highest levels of GAT1 expression with proline as nitrogen source (Fig. 7). This linear relationship is not observed with glutamine as nitrogen source until GAL1,10-GAT1-lacZ expression was much higher (Fig. 7B). A similar relationship was also observed with ammonia (Fig. 7C). These data are expected of a situation in which functional Gat1p is "titrating" an inhibitor with ammonia or glutamine as nitrogen sources, and increased DAL80-lacZ expression is not observed until Gat1p excedes a critical level, thereafter overcoming inhibition. Data in Fig. 5 suggest that the molecule Gat1p is likely titrating is Ure2p, because the effects of increased GAT1 expression are suppressed when URE2 expression is simultaneously increased. These results are consistent with the possibility that Gat1p forms a complex FIG. 6. A-J, ADH1-EGFP-GAT1 expressed in RTCY57 that overexpresses GAL1,10-URE2. K-Y, ADH1-EGFP-GAT1 expressed in wild type GYC86. Panels a, d, and g were stained with 4,6-diamidino-2-phenylindole; EGFP-GAT1p was visualized in panels b, e, and h. In panels c, f, and i, the 4,6-diamidino-2-phenylindole-positive material in panels a, d, and g was pseudocolored red, and the images were superimposed on those in panels b, e, and h.
with Ure2p as has been suggested for Gln3p (27).
Microscopic evidence extended the relationship between Gat1p and Ure2p by demonstrating that Gat1p is localizes predominantly to the nucleus when Gat1p-dependent transcription is occurring and is cytoplasmic when this transcription is inhibited by overproduction of Ure2p. In other words, Ure2p prevents Gat1p from reaching its physiological target, the GATA elements in NCR-sensitive gene promoters. A similar conclusion was reached for control of Gln3p operation from analysis of CAN1 expression. 2 Taken together, the data lead us to suggest that expression of GATA factor-regulated genes depends upon binding of the transcriptional activators Gln3p and Gat1p to their promoter targets and that once bound the activators are able to mediate transcriptional activation and gene expression. Although a negative observation, the finding that Gln3p tethered to DNA by LexAp mediates reporter gene transcription that exhibits little if any NCR-sensitive control (29) is consistent with this proposal. From this perspective, GATA factor regulation is achieved by regulating its binding to its promoter targets. At a coarse level this most likely occurs by regulating Gln3p and Gat1p access to the nucleus that is regulated by Ure2p. At a fine level, once the transcriptional activators have gained access to the nucleus, their overall level of operation is regulated by competition with the transcriptional repressors Dal80p and Deh1p for DNA binding. The fluidity and fine nature of the overall regulation is then achieved by the fact that production of three of the four GATA factors is autogenously and cross-regulated.
Data in Fig. 5B highlight an important characteristic of Ure2p participation in the NCR regulatory cascade. Increasing the intracellular concentration of Ure2p appears to eliminate the need for the physiological signal that intracellular nitrogen is in excess. This observation is consistent with the ideas that: (i) Ure2p exists as an inactive form that is activated in response to a signal indicating nitrogen excess and (ii) Ure2p exists in an active form that is inactivated in response to a signal indicating nitrogen limitation. By this reasoning, signal(s) generated in response to intracellular nitrogen excess or limitation can be circumvented by increasing the concentration of Ure2p or neutralizing negative control exerted by Ure2p by increasing the concentration of at least one of the GATA activators, in this case Gat1p. The simplest basis for such concentration-dependent control is protein-protein complex formation among the constituents of the regulatory circuit.
As this manuscript was being written, four reports simultaneously appeared reaching conclusions to which ours are both similar and complementary (30 -33). Although there is not full agreement on the mechanistic details, all propose that Ure2p complexes with one form or another of Gln3p, thereby preventing its entry into the nucleus when cells are grown in rich medium, and one of them demonstrated the proposed Gln3p nuclear and cytoplasmic localization (30). A similar model was proposed for Gat1p, although the investigators were unable to show any interaction between Gat1p and Ure2p. Our work points to such an interaction and contributes to filling this missing link in the proposed models.